Oestrogen and epidermal growth factor down-regulate erbB-2 oncogene protein expression in breast cancer cells by different mechanisms.

Mitogen-induced mammary cell growth is often accompanied by decreased levels of expression of the p185erbB-2 protein. We have previously reported that oestrogen inhibits erbB-2 mRNA and protein expression in breast cancer cells, while epidermal growth factor (EGF) treatment has been shown to decrease p185erbB-2 levels in normal mouse mammary epithelial cells. In the present work, we studied the effect of oestrogen and EGF on erbB-2 expression in oestrogen-responsive breast cancer cells. We observed that both oestrogen and EGF comparably down-regulated p185erbB-2 levels, while stimulating growth of T47D and ZR75.1 cells. Oestrogens, but not EGF, concomitantly down-regulated erbB-2 mRNA. Run-on analysis showed a reduced erbB-2 transcription rate in the presence of oestrogens. Furthermore, the transcriptional activity of a 219 bp proximal fragment of the human erbB-2 promoter was repressed by oestrogens, whereas it was enhanced by EGF. EGF stimulated both tyrosine phosphorylation and autokinase activity of p185erbB-2 down-regulates p185erbB-2 at a post-translational level. Thus, two factors converging in terms of effects on cell growth, display divergent mechanisms of regulation of erbB-2 expression.

ZR75. 1 cells. Oestrogens, but not EGF, concomitantly down-regulated erbB-2 mRNA. Run-on analysis showed a reduced erbB-2 transcription rate in the presence of oestrogens. Furthermore, the transcriptional activity of a 219 bp proximal fragment of the human erbB-2 promoter was repressed by oestrogens, whereas it was enhanced by EGF. EGF stimulated both tyrosine phosphorylation and autokinase activity of p185erbB-2 We conclude that oestrogens, but not EGF, inhibit erbB-2 expression by transcriptional repression, while EGF down-regulates pl85elbB2 at a post-translational level. Thus, two factors converging in terms of effects on cell growth, display divergent mechanisms of regulation of erbB-2 expression.
The erbB-2 oncogene encodes a 185 kDa tyrosine kinase receptor (pl85erbB-2) with homology to the EGF receptor (EGFR) (Yamamoto et al., 1986). p185erbB2, most likely in heterodimeric arrangements with the proteins encoded by the related genes erbB-3 and erbB-4 (Plowman et al., 1993;Sliwkowski et al., 1994), may constitute a receptor for the recently described family of peptides named heregulins, NDF, GGF and ARIA (reviewed by Mudge, 1993). erbB-2 activation is frequent in human cancer. erbB-2 is amplified in 20-25% of primary breast tumours (Slamon et al., 1987;Berger et al., 1988;Adnane et al., 1989) and overexpression of p185erbB-2 generally correlates with unfavourable clinical outcome (reviewed by Perren, 1991;Hynes, 1993). In most cases, erbB-2 overexpression is due to gene amplification; however, the amount of erbB-2 mRNA or pl85erbB-2 measured in some primary tumours and breast cancer cell lines does not directly reflect erbB-2 gene copy number (Kraus et al., 1988;Hynes et al., 1989;King et al., 1989;Dati et al., 1991), implying the existence of mechanisms regulating erbB-2 expression. One of these mechanisms may be the expression of specific transcription factors. The OB2.1 factor, which footprints the human erbB-2 promoter, is found in breast cancer cells that overexpressbut not in those that do notthe erbB-2 gene (Hollywood & Hurst, 1993). Hormones may also play a role in the regulation of erbB-2 expression and function. Oestrogens and EGF are obvious candidates for such regulation, since oestrogen receptor (ER) and erbB-2 or EGFR expression are inversely correlated in breast carcinomas (Harris & Nicholson, 1988;Perren, 1991). ER' tumours are generally well differentiated and less invasive, whereas tumours overexpressing erbB-2 and/or EGFR are less differentiated and more aggressive.
We and other groups have demonstrated that 17p-oestradiol, which is strongly mitogenic for ER' breast cancer cells, inhibits erbB-2 expression at both the mRNA and protein level (Dati et al., 1990;Read et al., 1990;Warri et al., 1992).
In developing rat mammary gland tissues, p185e`bB-2 increases progressively during pregnancy up to the complete functional differentiation state (Dati et al., 1990). This can be reproduced in vitro using the mouse mammary epithelial cell line HCI1, in which p185erbB-2 level increases on confluence and during hormone-induced differentiation, whereas stimulation of cell growth by EGF is accompanied by down-regulation of pl85erbB2 (Kornilova et al., 1992). The sum of these observations led us to ask whether erbB-2 down-regulation is a general effect linked to the entry of mammary cells in the replicative phase.
In the present study, we have compared the molecular mechanisms by which 17P-oestradiol and EGF regulate erbB-2 expression in ER' breast cancer cells. Treatment with either EGF or oestradiol comparably stimulated growth and concomitantly reduced p 185erbB-2 level. However, oestrogen treatment led to a repression of erbB-2 transcription, while EGF was found to act primarily at the level of erbB-2 protein.

Materials and methods
Cell culture ZR75.1 cells were obtained from the American Type Culture Collection. The oestrogen-responsive T47D cell line was obtained from Dr Salomon (Bethesda). Cells were maintained at 37°C with a 5% carbon dioxide atmosphere, in Dulbecco's modified Eagle medium (DMEM) containing 5% (T47D) or 10% (ZR75.1) heat-inactivated fetal calf serum (FCS), 4 mM L-glutamine, 20 mM N-(2-hydroxyethyl) piperazine-N'-(ethanesulphonic acid) (HEPES) buffer pH7.4, 50 IU ml' penicillin and streptomycin. This medium is designated 'complete medium' (CM). A medium devoid of oestrogenic activity (SM) was prepared by adding 5% dextran-coated charcoal (DCC)-treated FCS to Eagle minimum essential medium with no phenol red, and other supplements as in CM. Treatments were done with 17poestradiol from a stock solution in ethanol (maximum ethanol concentration in media 0.001%). The epidermal growth factor was human recombinant EGF (Sigma), used at 10 ng ml-' unless otherwise specified, except for autokinase experiments, in which mouse submaxillary gland EGF was used at 400ng ml-'.
RNA elongation assay Run-on assays were carried out essentially as previously described (Greenberg et al., 1986). Cells were homogenised in Dounce in buffer H (0.25 M sucrose, 10 mM magnesium chloride, 2 mM DTT, 0.1% Nonidet P40 and 10 mM HEPES, pH 8.0) on ice. Lysate was centrifuged at 600 g for 5 min and the pellet washed once:-The pellet was resuspended gently in buffer H containing 1 M sucrose, and centrifuged as above. Mg of the appropriate linearised and denatured plasmids had been transferred with a slot-blot apparatus. Hybridisation was done at 42°C in 50% formamide for 60 h. Strips were washed twice for 30 min in 2 x SSC at 65°C, then digested with 10 ,.g ml-' RNAse A for 30 min at 37°C, washed again in 2 x SSC, 0.1% SDS, for 1 h at 37°C and finally exposed to Kodak XAR films for 3-10 days at -80°C. Probes used were human erbB-2 cDNA in two different vectors, pSV2 and pLTR (DiFiore et al., 1987), the human c-MYC fragment pRyc7.4 (Nishikura et al., 1983), the ribosomal protein rpL7a cDNA, obtained from Dr S. Kozma (FMI, Basle, Switzerland) and sheared total genomic DNA from human placenta.
Reporter plasmids and chloramphenicol acetyl transferase (CAT) assay The human erbB-2 promoter fragment was derived from a genomic clone spanning more than 8 kbp on the 5' end of the gene. Cloning and characterisation of this clone, as well as construction of CAT reporter vectors, are to be published elsewhere. Part of the clone was sequenced and found to correspond to the erbB-2 promoter sequence previously published (Hudson et al., 1990). The construct used here (pE2P.PP.CAT) was composed of a 219 bp fragment, extending from the PstI site, located at position -397, to the major and most proximal transcriptional starting site, located at position -178, relative to the initiator codon ATG (Tal et al., 1987). Here, a BamHI site was introduced to allow cloning between the PstI and BamHI sites of plasmid pBLCAT3 (Luckow & Schutz, 1987). A human ER expression vector (pHEO) was obtained from Dr Chambon, Strasbourg, France (Green et al., 1986). Transient transfections were performed with 151g of pE2P.PP.CAT and 2 jig of pHEO, by the standard calcium-phosphate co-precipitation procedure. As a control for promoter specificity, parallel experiments were run with pRSV. CAT (Gorman et al., 1982).
Treatments were directly included in the medium used for transfections and renewed after 24 h. Cells were harvested 40 h after transfection by scraping on ice in PBS. Evaluation of CAT activity was performed by the thin-layer chromatography (TLC) method as previously described (Sambrook et al., 1989). Individual spots were cut and P-counted. A control for transfection efficiency was provided by co-transfecting 3 lsg of the P-galactosidase expression vector pCHI 10. p-Galactosidase was evaluated in cell lysates by the colourimetric method (Sambrook et al., 1989).
Tyrosine phosphorylation assay Cells were lysed on ice in PT buffer containing 20 mM sodium molybdate and 20 mM sodium fluoride and the lysate cleared by centrifugation at 2,000 g for O min at 4°C. pl85erbB-2 was immunoprecipitated at 4°C with the 21N antiserum and collected on protein G-Sepharose. Beads were washed in ice-cold PT buffer, boiled for 5 min in SDS-PAGE buffer, the denatured proteins resolved on 8% PAA gels and gels blotted onto PVDF filters. Filters were incubated with erb-B2 REGULATION BY OESTROGEN AND EGF IN BREAST CANCER CELLS the 4G10 anti-phosphotyrosine MAb (UBI) and revealed by the ECL method as described above. To provide a control for the amount of p185 in the lysate, filters were stripped and reprobed with the 21N antiserum.

Results
Effects of oestrogen and EGF on erbB-2 mRNA and protein expression The effects of oestrogen and EGF on erbB-2 expression were studied using the two ER' breast cancer cell lines T47D and ZR75.1, in which erbB-2 expression and regulation was previously characterised (Dati et al., 1990;Antoniotti et al., 1992;Taverna et al., 1994). T47D and ZR75.1 express moderate levels of both pl85ebB-2 and EGFR as compared with other breast cancer cell lines. They show 5-to 10-fold less pl85erbB-2 than cells with amplified erbB-2, such as SKBR.3 (Kraus et al., 1988;Hynes et al., 1989) and express 5-10 x I03 EGFR sites per cell (Koga et al., 1990;M. De Bortoli, unpublished results). Moreover, they show comparable sensitivity to 17poestradiol in terms of cell growth.
The effects of 17p-oestradiol or EGF treatment on the expression of erbB-2 mRNA and p185erll-2 were evaluated by, respectively, Northern and Western blotting. Figure 1 shows a comparison between the levels of pl85elbB-2 and of erbB-2 mRNA in ZR75.1 cells after 2 days of treatment with the effectors. In these conditions, cell growth in the presence of 17p-oestradiol and EGF was, respectively, 281 ± 64% and 216 ± 108% of the growth in the absence of the factors, as determined by direct counting of viable cells. Both 17,oestradiol and EGF induced a dramatic decrease in pl85erbB2. However, a parallel decrease in erbB-2 mRNA was observed only in 17p-oestradiol-treated cells. EGF-treated cells contained 30-40% more erbB-2 mRNA than control cells, as determined by densitometric evaluation of the blots. Downregulation of p185er8B-2 by EGF was roughly equivalent whether the treatment was carried out in complete medium (CM) or in medium with charcoal-treated FCS (SM). The effects of oestrogens on both growth and erbB-2 expression are evidenced only when cells are cultured in SM, i.e. in the absence of oestrogen. Culture of the cells in SM brings about a progressive increase in erbB-2 protein level as well as promoter activity during 2 weeks of culture (Dati et al., 1990;Taverna et al., 1994). Reprobing of the protein blots with the OD3 monoclonal antibody, which recognises the external domain of human p185erbB-2, gave similar results (not shown). This ruled out the possibility that EGF effect might result from masking of the C-terminus of pl85erbB2, the epitope recognised by the 21N antibody used throughout this study.
The effects of 17P-oestradiol and EGF on pI85erbB-2 were dose dependent. As shown in Figure 2, after culturing T47D cells in SM for 4 days and treating for additional 4 days, maximal effect was seen with 1 x 10-9M and 1 x 10-8M 17p-oestradiol; the EGF effect was clearly seen at concentrations as low as 1 ng ml', corresponding to 1.7 x 10-1"M.
These values are consistent with receptor-mediated effects: the affinities of ER for 17p-oestradiol and of the EGFR for EGF in T47D cells are respectively: Kd-0.5 x 10' and Kd-1 x 10-9 M (Koga et al., 1990;M. De Bortoli, unpublished).
pl85erbB-2 down-regulation by both 17p-oestradiol and EGF was relatively slow. Reduction of p185erbB-2 was seen after 5 h of treatment (Figure 2, right), but a time-course analysis showed that the maximal response to both factors occurred after 4-5 days of treatment (not shown). However, it is important that cells do not reach confluence during this time, since up-regulation of erbB-2 expression by cell confluence takes place (Taverna et al., 1994). For this reason, in all the experiments described here, conditions were set in order to avoid reaching confluence degrees higher than 60-70%. Similar dose and time dependence were measured on ZR75.1 cells (not shown).
EGF induces tyrosine phosphorylation ofpJ85`erbB2 in ER' breast cancer cells It has been reported that EGF induces pl85erbB-2 phosphorylation Stern & Kamps, 1988). We examined this in ER' breast cancer cells that express moderate levels of both pl85erbB-2 and EGFR. First, we studied the in vitro kinase activity of pl85erbB-2 following EGF treatment. Treatment of T47D cells with EGF at 37°C for 10 min caused a 3.3-fold increase in the kinase activity that co-immunoprecipitated with pl85erbB-2, as determined by quantitation with a phosphoimager (Figure 3). A second, lower band is visible in EGF-stimulated cells. Since heterodimerisation of the EGFR with pl85erbB-2 has been reported in SKBR3 breast cancer cells (Goldman et al., 1990), this band may be due to co-immunoprecipitated EGFR.
Tyrosine phosphorylation of pl85erbB-2 after EGF treatment of T47D cells was studied by blotting with an antiphosphotyrosine antibody. Immunoprecipitated p185erbB-2 from EGF-treated cells, but not from control cells, contained phosphotyrosine, as shown in Figure 4. EGF-induced pl85erbB-2 phosphorylation was transient, being maximal at 5 min and decreasing thereafter. Reprobing of the blot with the 21N anti-pI85erbB-2 antibody provided a control for the amount of p185erbB-2 in each lane (Figure 4, bottom). As in Figure 3, the lower tyrosine-phosphorylated protein which co-immunoprecipitated with p185erbB-2 may represent EGFR.   Kinase activity of pl85erbB-2 following EGF treatment of T47D cells. Cells were preconditioned in 1% FCS-DMEM for 24 h, then treated or not with 400 ng ml-' mouse EGF for 10min at 37°C. pl85erbB-2 was immunoprecipitated with either a preimmune (PI) or the FRP5 anti-p185erbB-2 monoclonal antibody on anti-mouse IgG-coated protein A-Sepharose. Beads were incubated with [y-32P]ATP for 10 min at room temperature, then proteins were solubilised and separated on 8% polyacrylamide-SDS gels.
Similar results were obtained on ZR75. 1 cells (not shown). When this work was in progress, it was reported that in fibroblasts not expressing the ER and stably transfected with human erbB-2 there was a rapid and transient activation of pl85erbB-2 by 1 x 10-6 M 17P-oestradiol (Matsuda et al., 1993).
For this reason, we examined the effects of either a physiological concentration (1 x 10-8 M) or a non-physiological concentration (1 x 106 M) of 17p-oestradiol on erbB-2 tyrosine phosphorylation in T47D cells. Figure 4 shows that a 10 min treatment with both doses of 17p-oestradiol led to a slight increase in p185erbB-2 phosphotyrosine, as compared with cells cultured in SM without oestrogens. The effect of oestrogen was more than 10-fold less than that evoked by EGF in similar conditions, These data demonstrate that in ER' breast cancer cells with low levels of both receptors EGF induces tyrosine phosphyorylation of pl85erbB-2, most likely through its own receptor, while 17p-oestradiol, even at very high concentration, does not lead to the same extent of pl85erbB-2 phosphorylation as that induced by EGF.
Oestrogens inhibit erbB-2 expression at the transcriptional level We investigated whether the oestrogen effect on erbB-2 mRNA might be due to inhibition of erbB-2 gene transcrip- Anti-p1 85 w tion. The numbers of transcripts initiated in the presence and in the absence of 17p-oestradiol by nuclei of ZR75.1 cells were evaluated by a nuclear run-on assay. Figure 5 shows the results obtained on cells treated with or without 17P-oestradiol for 48 h. Quantitation of hybridised slots showed that oestrogen-treated cells possess about 50% of the initiated transcripts of erbB-2 compared with control cells. No change in c-MYC gene transcription following oestrogen treatment was seen, in keeping with the fact that long-term enhancement of c-MYC expression by oestrogen in breast cancer cells is due to stabilisation of the c-MYC mRNA (Santos et al., 1988). The effect of EGF on erbB-2 transcription was not studied.
Since this experiment demonstrated transcriptional inhibition of erbB-2 by oestrogens, we asked whether the elements mediating transcriptional repression were located within the promoter of the erbB-2 gene or were located in more distal regions of the gene. In fact, examination of the published sequence of the first 1.2 kbp of the human erbB-2 promoter (Hudson et al., 1990) did not reveal any oestrogen-responsive consensus element. A CAT reporter plasmid, containing the most proximal 219 bp fragment of the human erbB-2 promoter, extending from the PstI site at -397 to the major transcriptional starting site at -178 relative to the initiator ATG, was derived from a human genomic clone. This construct, called pE2P.PP.CAT, was transiently transfected in T47D cells and expression of CAT after treatment with 17P-oestradiol and EGF evaluated. In Figure 6 an example of CAT assay on T47D cells treated with 17p-oestradiol or EGF is shown. Averaging three independent transfections in the same conditions, and normalising the activities on the basis of the co-transfected ,-galactosidase, it was possible to calculate a _ 60% repression by 17i-oestradiol and a 4to 5-fold stimulation by EGF.
Calculated values with standard deviations are shown in Figure 6. In order to observe transcriptional repression by 17fr-oestradiol, it was necessary to co-transfect the ER expression plasmid pHE0. In its absence, the transcriptional response was 3-to 4-fold less. No Figure 7 shows that CHX itself, in the absence of oestrogen, decreased erbB-2 mRNA by 41% (compare lanes e and j). In 24 h, 17,oestradiol induced a 57% decrease in erbB-2 mRNA in the absence, but not in the presence, of CHX (compare lanes d-e and i-j). As a control for CHX efficacy, the blot was rehybridised to a c-MYC probe. At 3 h, expression of c-MYC is transiently induced by 17p-oestradiol (lane a) and superinduced by CHX (lane f), as previously described (Greenberg et al., 1986). Although ethidium bromide staining of the gel and methylene blue staining of the filter revealed no significant differences in the amount of RNA present in each lane, minor differences were seen in both mRNAs used for comparison, i.e. P-actin and GAPDH. Evaluation of several similar experiments, which gave similar results, allowed us to conclude that inhibition of protein synthesis abolishes oestrogenic repression of erbB-2. ing the CAT gene. Similar results were obtained on ZR75.1 cells (data not shown). Finally, we examined whether repression of erbB-2 by oestrogen requires protein synthesis. This question was justified by both the absence of an oestrogen response element in the oestrogen-repressible fragment of the erbB-2 promoter and by the relatively slow kinetics of oestrogen action. ZR75. 1 cells precultured in SM for 4 days were treated with 17P-oestradiol for 3, 6, 12 and 24 h in the presence or absence of cycloheximide (CHX) and the levels of erbB-2 mRNA measured by Northern blotting. Semiquantitative estimation of erbB-2 mRNA was made by computerassisted image analysis. Values were corrected on the Cycloheximide (50 1M) was added I h before adding 17p- oestradiol.

Discussion
This study shows that oestrogen and EGF, two agents which stimulate growth of ER' human breast cancer cells, cause a down-regulation of pl85e"bB.2. The mechanisms leading to a decrease in p185erbB-2 differ. Despite the fact that EGF has a positive effect upon erbB-2 mRNA levels, its effects upon the kinase activity and tyrosine phosphorylation of pl85erbB-2 lead to decreased protein levels. Conversely, p185erbB-2 downregulation by oestrogens is accompanied by a parallel decrease in erbB-2 mRNA, transcription rate and promoter activity. It is known that EGF can induce pl85ebB-2 phosphorylation on tyrosine (Stern & Kamps, 1988;King et al., 1988). The mechanism of transphosphorylation involves receptor heterodimerisation (Wada et al., 1990) and heterodimers of pl85erbB-2 and EGFR have been detected in the SKBR.3 breast cancer cells, which carry several copies of the erbB-2 gene and express high levels of p185erbB-2 (Goldman et al., 1990). Heterodimeric forms have higher affinity for EGF than EGFR homodimers and possibly activate separate transduction pathways (Wada et al., 1990). It is likely that heterodimerisation of EGFR and pl85erbB-2 also takes place in T47D cells, since in the in vitro kinase assays and in the phosphotyrosine blot a second band of lower molecular weight is visible, indicating the coimmunoprecipitation, with p 85erbB2, of a phosphorylated _ 170 kDa protein. In these cells it was not possible to obtain direct evidence of heterodimer formation, probably because of the low level of expression of both receptors. EGF treatment of HC 11 mouse mammary cells led to a down-regulation of p185erbB-2, by increasing its phosphorylation and accelerating the rate of its internalisation and degradation (Kornilova et al., 1992). It is possible that in T47D cells a similar mechanism takes place. Further studies may elucidate this possibility. We noticed a slight positive effect of EGF on the steadystate erbB-2 mRNA level and a strong positive effect on erbB-2 promoter activity in T47D cells. EGF has also been shown to stimulate an erbB-2 promoter-luciferase reporter gene in HeLa cells (Hudson et al., 1990). The effect of EGF on erbB-2 promoter activity is much stronger than its effect on the steady-state level of erbB-2 mRNA. This may reflect either post-transcriptional regulation or the presence of more distal repressors in the erbB-2 regulatory sequences. Indeed, an erbB-2 promoter construct extending to position -1,398 showed a very small response to EGF in both transient transfection on T47D and in stable T47D transfectants (Taverna et al., 1994). Studies are under way to localise the element(s) modulating these transcriptional responses to EGF in mammary cells.
Down-regulation of erbB-2 by oestrogens appears generally stronger at the protein level than at the mRNA level, a finding noted by others (Russel & Hung, 1992). In the experiments reported here we observed a 50-70% inhibition of erbB-2 transcriptional activity, which appears lower than the decrease in p185er`R-2 seen in immunoblots. As mentioned ca v ,, a e; r g n above, a rapid activation of pl85erbB-2 kinase by high concentrations of 17p-oestradiol, was observed in ERmouse fibroblasts expressing high levels of human erbB-2 (Matsuda et al., 1993). The authors suggest that 17P-oestradiol may directly bind to pl85erbB-2, leading to an increase in its kinase activity and its internalisation. These provocative findings await further experimental proof. However, a rapid and transient ER-dependent increase in the phosphotyrosine content of several cellular proteins, following oestrogen treatment of MCF7 breast cancer cells has also been reported (Migliaccio et al., 1993). Our data show that the extent of pl85erbB-2 tyrosine phosphorylation induced by oestrogens in T47D cells is slight, yet it may account for the different amplitude of response to oestrogens observed between erbB-2 protein and mRNA or transcription. Oestrogens clearly inhibit erbB-2 expression at the transcriptional level. Oestrogen decreased both the number of erbB-2 transcripts which can be elongated in vitro and transcription of the human erbB-2 promoter-CAT reporter gene. The degree of such inhibition is probably dependent upon the level of ER present in the cell, since co-transfecting an ER expression vector resulted in a much larger transcriptional repression (50-70% vs 20-30%). However, the endogenous ER present in T47D cells is sufficient to mediate a 60% repression of a stably integrated 1.2 kbp erbB-2 promoter-CAT construct (Taverna et al., 1994).
Transcriptional repression of the rat NEU gene by oestrogen-activated ER has been reported (Russel & Hung, 1992). The responsive region is contained within a 144bp fragment near, but not contiguous, to the transcriptional starting sites. The promoters of rat NEU and human erbB-2 are nearly identical in their proximal portion (White & Hung, 1992); the oestrogen-repressible rat NEU 144 bp fragment corresponds to positions -354 to -210 of the human erbB-2 promoter, and thus it is contained entirely in the construct used here. Importantly, this fragment contains the two most proximal of the three footprints revealed in T47D and ZR75.1 cells (Hollywood & Hurst, 1993). Sequence analysis of this fragment reveals several potential regulatory elements, including OTFI, MYB, Spl, K-enhancer, which are contained in the footprinted sequences. In addition, this fragment was shown to be repressed by the MYC and ElA oncogenes (Suen & Hung, 1991). Repression of erbB-2 promoter-CAT constructs by cotransfection of a c-MYB expression vector has also been observed (P. Maggiora et al., in preparation). Both c-MYC and c-MYB expression is stimulated by oestrogens in breast cancer cells (Santos et al., 1988;Dati et al., 1990;Collyn d'Hooge et al., 1991), which leads to the hypothesis that the c-myc and/or c-myb proteins may mediate the oestrogeninduced repression of erbB-2 transcription.
The question of whether oestrogen may require protein synthesis to inhibit erbB-2 was approached by using cycloheximide. It must be emphasised that, given the slow kinetics of erbB-2 regulation and the short half-life of the oestrogen receptor (see below), it is extremely difficult to interpret these results. As already mentioned, during the growth of T47D or ZR75.1 cells in SM, i.e. in medium deprived of steroids, the expression of both erbB-2 mRNA and pl85erbB-2 increases for several days (Dati et al., 1990;Read et al., 1990;Russel & Hung, 1992;Taverna et al., 1994). The treatment with CHX indeed blocked not only the inhibitory effect of oestrogen but also the increase in erbB-2 expression in SM. One explanation for this is that the increasing expression of erbB-2 in SM may reflect an increased synthesis of trans-activator(s), and that inhibition of erbB-2 expression by oestrogen may be due to trans-repressor(s) synthesis, e.g. c-MYB or c-MYC, as discussed above. However, it is clear that, in the absence of protein synthesis, the amount of oestrogen receptor itself is significantly reduced, as a half-life of 3-5 h has been reported in MCF7 breast cancer cells in either the presence or absence of oestrogens (Eckert et al., 1984). Reduction of oestrogen response by CHX may then simply reflect the progressive disappearance of oestrogen receptor. Direct repression of gene expression by members of the steroid receptor family may be exerted through titration of positive co-factors or by competition with positive factors (for review, see Beato, 1991). Such a mechanism would be compatible with our observations. Further studies are under way to understand the exact molecular mechanism of erbB-2 repression by oestrogen.
In conclusion, the data presented here show that there is an important difference between oestrogen-and EGF-stimulated breast cancer cell growth. Since oestrogens lead to a loss of p185erbB-2, signal transduction through p185erbB-2 homodimers or heterodimers with other members of the EGFR family is abrogated. In contrast, EGF treatment leads to the activation of pl85ebB-2. These differences may be very significant and partially explain the different behaviour of ER' and EGFR+ breast tumours.